Agentic Sandbox Escape Proves Sandboxing Isn’t Enough
The consensus take on agentic sandbox escape is simple enough: a powerful model was told to break out, it did,…
-
-
AI Agents Lied to Sponsors: And That’s the Point
The Manchester story about AI agents sounds like a joke until you notice what actually happened. Three developers gave an…
-
Video Object Removal: What VOID Gets Right
The demos of Netflix’s new VOID model make video object removal look like sorcery: delete a person and the guitar…
-
ChatGPT Extension Privacy: The Browser Itself Is the Leak
A Chrome extension with a shiny “Featured” badge quietly scraped ChatGPT and DeepSeek conversations from roughly 900,000 browsers, batched them…
-
Claude Code Leak: Why the Harness, Not the Model
If you tried to clone Claude Code last week, the hard part wasn’t the model. It was rebuilding half a…
-
Anthropic Mythos: Step Change, Power, and Risk
A few years ago, frontier AI felt like a software story: new models every quarter, bigger contexts, better alignment, mostly…
-
Anthropic Data Leak: How Ops Failures Undermine AI Safety
Anyone with a browser and a bit of curiosity could quietly pull draft pages about Anthropic’s unreleased “Claude Mythos” model,…
-
LiteLLM PyPI Compromise: Thin Wrapper Steals Keys
A single pip install of LiteLLM 1.82.8 was enough to run a credential stealer every time Python started, thanks to…
-
Pentagon Palantir AI: Why Locking Maven Matters Now
Palantir’s Maven Smart System just graduated from “AI experiment” to “the way the U.S. military fights wars.” A Pentagon memo…
